How our PR firm is using AI and how you can too with Jason Mudd, CEO of Axia Public Relations

On Top of PR podcast: AI in PR with show host Jason Mudd episode graphic

In this solocast, host Jason Mudd describes how our PR firm is utilizing AI, as well as how you can use it to benefit your own company. He discusses the five most important things to remember when working with AI in your agency.

Tune in to learn more!

Watch the episode here:

5 things you’ll learn during the full episode:

Starting to use AI
Using caution with AI
Setting up policies around using AI
Communicating the policies
Enforcing the policies

Resources

Find Jason Mudd on Twitter.
Connect and learn more about Jason Mudd on LinkedIn.

Additional Resources from Axia Public Relations:

Listen to more episodes of the On Top of PR podcast.
Find out more about Axia Public Relations.
Create a usage policy for AI-generated content.
How PR Professionals Can Use Artificial Intelligence Tools To Enhance Their Work
ChatGPT

Episode Highlights

[02:55] Starting using AI

Use AI. Don’t be on the sidelines watching everyone else adopt a new technology.

Jason: “And like anything else — when mobile technology came out, when social media came out, when e-commerce came out — you can either ignore those things or you can be early adopters of those things and start leading the way and pioneering the way.”

Jason: “You've got those that are afraid of it or saying, making comments like, ‘Why would you even use it? All you're doing is giving the beast the information it needs to make us obsolete.’ There are people who just think it's unprofessional and kind of amateur to use AI. And then I felt like there was another group of people that were very focused on AI and using AI, and I think we lean much closer to that group, realizing that AI is a tool that can help you be more productive.”

[06:47] Using caution with AI

The number-one thing to do is think about ethics.

Jason: “When you think about ethics, you have to think of things like disclosure, You have to think of things like copyright, citing sources, fact-checking, the confidentiality of the input or the information that you're putting into the machine or into the platform or the system, and work product ownership.”

Neither you nor AI can own the work that is generated since it is collaborative work. You can change the content and draw on ideas from the generated work and then it is a different case. The best advice is to involve your legal counsel in these decisions.

Jason: “When it comes to confidentiality, though, think about formal nouns and perhaps omitting formal nouns or inserting alternate formal nouns into your input. So in other words, instead of saying Axia Public Relations or Jason Mudd, or Coca-Cola, you could just put in an acronym or another name that is different.”

Consider the possibility of a data breach and ensure you don’t put confidential information into the system.

It is also important to ensure that the information is accurate, correct, and current. Fact-checking the information it generates is an essential step in using AI.

[11:07] Communicating and enforcing policies around AI

Jason: “You really are going to have to take this [AI] rough draft, and I emphasize the rough in it, and finesse it, tweak it, and modify it to get it to where it needs to be production ready.”

AI isn’t always perfect, so it needs a human touch to make it ready to be seen and interacted with by the public.

AI can be used for inspiration and a general idea of what the final product could look like.

[12:27] Ways to use AI

Jason: “We are not using AI to replace the talent that we have at our agency or the talent assigned to your account. Instead, we’re using AI to keep those people doing what is called their highest and best use — HBU.”

The highest and best use is a real estate term generally describing the idea of maximum productivity.

About Jason Mudd

Jason Mudd is a trusted adviser and dynamic strategist for some of America’s most admired brands and fastest-growing companies. Since 1994, he’s worked with American Airlines, Budweiser, Dave & Buster’s, H&R Block, Hilton, HP, Miller Lite, New York Life, Pizza Hut, Southern Comfort, and Verizon. He founded Axia Public Relations in July 2002. Forbes named Axia as one of America’s Best PR Agencies.

Enjoy the Podcast?

If you did, be sure to subscribe and share it with your friends!

Post a review and share it! If you enjoyed tuning in, leave us a review. You can also share this with your friends and family. This episode can give you professional insight into media coverage. Know your rights and the regulations to follow when it comes to the media.

Have any questions? You can connect with us on Facebook, Twitter, and LinkedIn.

Thank you for tuning in! For more updates, visit our On Top of PR website or join the community. You may also tune in on Apple Podcasts, Spotify, and YouTube.

Transcript

[Announcer]

Welcome to On Top of PR with Jason Mudd presented by ReviewMaxer.

[Jason]

Hello and welcome to On Top of PR. I’m your host, Jason Mudd with Axia Public Relations. It's been a while since we recorded an episode. We have been busy traveling with clients at conferences and other types of activities. So we kind of informally, unofficially took the month of May off without recording any episodes.

So, here we are recording a solocast where it's just going to be you and me. And so welcome back. Thank you for your patience and loyalty. We are glad to be here. Glad to be back sharing some insights with you as part of our On Top of PR program. We’re more than 100 episodes in and couldn't do this, and wouldn't want to do this, without you and your support and participation. So thank you for that. Today's a solocast where it's just you and me talking and I'm sharing some of the things that I've learned or observed or I think is important and timely for you to be aware of.

At the conferences that we went to, there was a lot of chatter about artificial intelligence, and you could tell there's kind of different camps when it comes to AI. You've got those that are afraid of it or saying, making comments like, ‘Why would you even use it? All you're doing is giving the beast the information it needs to make us obsolete.’ There are people who just think it's unprofessional and kind of amateur to use AI. And then I felt like there was another group of people that were very focused on AI and using AI, and I think we lean much closer to that group, realizing that AI is a tool that can help you be more productive.

And like anything else — when mobile technology came out, when social media came out, when e-commerce came out — you can either ignore those things or you can be early adopters of those things and start leading the way and pioneering the way. So that's what I want to talk to you a little bit today is about artificial intelligence. When I mention AI tools or I mention ChatGPT specifically by name, please keep in mind that that's one of many tools that are out there that you could be utilizing for AI in your workplace and your PR department or your corporate comms department, or like us at our PR firm and PR agency.

So, I've already kind of gotten ahead of myself a little bit. But the five things I want to talk to you about today was one: use it, don't be on the sidelines. So take the opportunity to decide, ‘Are we going to use AI?’ And the next step is ‘How are we going to use it?’ So step two is kind of set up policies, right? Set up the guardrails, and there's a whole lot to be mindful of. So before we do that, actually, let's talk about using caution, which is one of our five steps today. So using caution when you're thinking about using artificial intelligence, so there are so many things to be thinking about. So if you've got something to write with and it's handy, let's use that here as we are talking about these things right now.

So, number one, ethics — the first thing you should always think about is ethics when you're looking at a new program, a new offering, a new idea, a new strategy, a new message, a new direction. What is the ethical behavior of using AI in your workplace? What are the ethical concerns that you need to think about? Is that giving attribution to say that this content was developed by AI, that this content was, you know, we used AI to produce this communication or whatever it might be?

And so obviously, when you think about ethics, you have to think of things like disclosure. You have to think of things like copyright, citing sources, fact-checking, the confidentiality of the input or the information that you're putting into the machine or into the platform or the system, work product ownership. Right now, the common legal opinion is that you can't own the work product that is developed by AI, that the AI engine itself doesn't really own it either, because it's a collaborative process of you inputting the information and the AI system outputting information. And so you probably want to involve your legal counsel in those discussions.

When it comes to confidentiality, though, think about formal nouns and perhaps omitting formal nouns or inserting alternate formal nouns into your input. So in other words, instead of saying Axia Public Relations or Jason Mudd, or Coca-Cola, you could just put in an acronym or another name that is different. But you know what that name is, so like an alias that you could use instead because the question becomes, is it confidential, the information that you put into the system, and what happens if there's ever a data breach? So be mindful of that. And then also use caution with the inputs or the commands that you give to the AI because I find that oftentimes the commands we're giving it are not very sophisticated. They're not as specific as we want them to be. And so the nice thing is you can always go back in and you can improve your input.

And there are several other factors, including one I just discovered last week, which is I was asking it to do something for me and it came back and said, ‘Hey, this is the best I can do because my knowledge cutoff date is September 2021. And so I don't have any new information in my system since September 2021.’ Now, I don't know if the AI tool, which by the way, this was ChatGPT at the time, and if you know the answer, please comment back to us about this. But I don't know if it was specifically saying I don't have any new information about this particular topic or if its knowledge cutoff date is September 2021 across the board. That was my perception based on what I saw and the way it presented it to me. So that's very concerning because at this time of recording, it's now June 2023, which means we're getting awfully close to September, which means that the input is 18 to 20 or maybe even closer to 24 months old, and the world moves fast and data changes and content changes and news happens every day. So, I was a little concerned when I saw that piece of information, but I'm glad it disclosed it to me instead of giving me poor, incorrect, or inaccurate information.

So, step one is to use it, right. Step two is to use caution. Step three is to set up policies. And so we talk, you know, the ways to do that is to consider the caution that I gave you: the ethics, the disclosure, copyright sourcing, citing sources, fact-checking, confidentiality, work product ownership, input specialization and techniques, knowledge cutoff dates, and other factors.

And then once you set these policies, it should probably look like an employee policy. You should probably create a vendor policy of how you want your vendors using and not using it. So you got an employee policy, a vendor policy, and maybe you need other policies for different departments or different roles in the organizations. And then step three is you need to communicate about these policies both internally as well as externally.

And what are the ethics of communicating these policies? So do you add this to your employee handbook? Do you have people sign a pledge or a commitment or a code of ethics document specifically to how to use AI? And then for your external audiences, are you adding these to your customer agreements? Are you notifying customers in the terms and conditions or in some sort of proactive communication? Maybe you have clients and you want your clients to know that you're using AI to give them a better solution or to generate more ideas.

I guess the other thing I should talk about under use caution, and I hate to back up a little bit, but as I'm sharing this information with you, I'm reminded that I should probably set the stage and let you know that to me, AI works really fast and is a very helpful and powerful tool. But the issue is the quality of the tool. The quality of the tool reminds me of an intern, a college student, maybe even a high school student as far as the quality of the work goes. But that doesn't mean that the final product AI produces is ever going to be close to your final product.

And I think where the challenge will be is there are people who are going to resist AI, and there are going to be people who will not be able to leverage AI because let's say, for example, and I'm not thinking of anybody in particular, but someone may not be a very good writer. They turn to an AI tool like ChatGPT to write a first draft and perhaps that first draft ChatGPT wrote, is as good as the person giving the input could have written it themselves. And for that situation, I'm concerned because what we're doing at our agency is we are getting that first draft from an AI tool. And by the way, we use about five or six at this point because a lot of our vendors, our SAS, and our service bureaus are already providing AI tools embedded within the services that we're using from them.

The point is, if let's say you're writing is, let's just say a C quality already, then it'll be hard for you to get it to be a B quality. But if you're like the professionals I think we employ, they write A-level work all the time and so ChatGPT or some other AI-embedded tool can give us a C-level first draft. Then we have to spend time on improving that draft from a C to a B, from a B to an A, right? We also have to fact-check that document. We also have to make sure there’s no copyright issues. We have to also make sure that, there are other challenges, the sources are cited accurately and things like that.

So, there are some cases where I would say maybe it's helpful just to get some ideas from the engine or from the tool and then write yourself, because I think you could spend a whole lot of time taking a first draft from AI, especially where it is right now, and turn that into something that's production ready, that's ready to go live, that's ready to be seen by the public or your client or your employer. So, be thinking about that. But you really are going to have to take this rough draft, and I emphasize the rough in it, and finesse it and tweak it and modify it to get it to where it needs to be production ready.

Okay. So going back to you communicate about these policies, but then you've got to enforce these policies, and the way to enforce it is there are tools out there that will tell you what percentage of this document it perceives to be AI-generated. You also can use tools like plagiarizing awareness tools. So tools that compare the copy that you're putting in front of and say, ‘Hey, is this plagiarized?’ And it goes out and searches the web and tries to see what that looks like. And what that will also help you do is identify how much of this is what's called duplicate content, or content that the search engines will see, ‘Gosh, this content is already out there on this other website. It was there first. That website has a higher domain authority. So we're not going to issue or we're not going to rank this page high or at all because it's considered to be duplicate content.’ There's ways around fixing duplicate content, there's ways around proper attribution and linkback from duplicate content to avoid those things.

But again, you're going to need to figure out a way to enforce the policy that you've created. So like Ronald Reagan said, ‘trust but verify’ — so you need to verify that people are following the policies that you've created and you need to, of course, enforce those policies.

Now, I want to just share, as we're wrapping up here, I want to share some of the ways that we have used AI within the past week. This is not going to be a complete and comprehensive list. And some of these things may have been done more than a week ago, but I was just trying to collect some thoughts as ways I've heard or I'm aware that we're using AI in our agency. And again, we are not using AI to replace the talent that we have at our agency or the talent assigned to your account. Instead, we're using AI to keep those people doing what is called their highest and best use, HBU. That's a common term in real estate, highest and best use. And so, to that end, having a first draft written to inspire or some guidelines or an outline is certainly helpful. Then we spend a lot of time tweaking and massaging and perfecting and tightening up the language and fact-checking and making sure there's proper attribution, all that stuff.

But if we can get a first draft in front of us very quickly, sometimes that's helpful, sometimes it's not, but it's worth a try. And I will tell you, just the other day I was using AI and I said, ‘Hey, give me five compelling reasons for X or tell me five things about Y.’ And those were all types of information and directions I wouldn't have gone in and content I probably wouldn't have thought of that helped improve, helped the work that I was already doing.

So, just the other day, for example, we were working on our own company policy. We have one. It's written, it just hasn't been disseminated. So we're working on doing that and finalizing that. We're going to have people sign something, a statement about their pledge to use AI in an ethical way. So we have a policy already created.

But then I thought, let's go into AI and see what kind of policy AI comes back with, and then let's find a way to merge the two. So we've got an even better end-result product. So that's a perfect example. You did the work already on your own, but then you went to AI to borrow from some other angles, topics, ideas, and recommendations that you may or may not end up ultimately using.

Okay, so the next thing we did is once we put input the request to create a company policy, then we said, ‘Okay, draft an email from the company leadership, the H.R. department, whatever it might be, announcing that we've created a company policy and that they will be receiving it from the company soon.’ And then we asked it to create a cover letter to go with the company policy.

And then we just started getting creative and having some fun with it and saying, ‘Okay, let's turn this into a news release announcing that we've created a policy. We'll probably not issue a news release about it, but it was just cool to see what I put together and what it thought. And maybe there was an idea in there that we hadn't thought of that then makes it interesting and compelling for us to go out and do a news release from that. The input we gave it was to say, ‘Okay, now put together a media pitch for this company policy announcement,’ and then obviously it begs the question or the opportunity to say, ‘Okay, what media lists should we use to send this out?’

So, we put that in there. And believe it or not, there are a couple of media outlets that we had never thought of. There are a couple of media outlets in that list, honestly, that we should never contact. And so again, it's a rough draft. It's an entry-level virtual assistant giving you some guidance is what I would describe it as.

And so, as we all know, entry-level folks come up with great ideas, but some of the ideas just aren't a fit, or aren't appropriate, or aren't sophisticated enough, aren't reliable enough, aren't, you know, knowledgeable enough to go live with. Now, honestly, would we ever just input these things into AI and just use them blindly? No, absolutely not.

That would be poor-quality, low-quality work. But again, it's a fast movement or acceleration to get you to that destination. So playing around more with these tools, you know, we've got a client and we asked them to write an executive bio for that person. And again, there are lots of issues with the bio. They kept calling the person a senior vice president, which that person has never held that title of senior vice president in their career.

So, again, still some issues, still some concerns, still some things to be thinking about. But there was content in there that was an interesting way of wording something or an interesting way of referencing something or an angle we might not have thought to take that we could consider taking in our own work product. Where I have found these tools to be very helpful and candidly fun and very quick is developing a product name, a company name, even a meeting name or a theme for a meeting kind of thing.

So, I've used this several times. There's a group of PR agency owners that I meet with once a month, and I want to come up with a cool, fun name for that time that we get together. And so I input some information, I used ChatGPT for this. I didn't use any formal nouns. I didn't put any real people's names or real agency names.

I just said, ‘Hey, got a group of PR firm owners. We meet once a month, here's what we call ourselves as a group. What could we call our group meetings?’ And then when it came back, I think it gave me 10 ideas. I told it, ‘I like idea three, seven, nine and ten. Come back with more like that.’ It came back with more ideas. I said, ‘Hey, this is good. I like where you're going with number one, number six, and number four. Come back with more like that. Then it came back and I said, ‘Okay, I like these two. Let's keep them to three words. No more than three words.’ So then it came back and did that.

So, then I said, ‘Okay, I like these. I like that. Now let's come back and make them alliterations. I think that would be fun.’ And I don't know the exact order of what came next, but at some point I said ‘Okay, make these punchier, make these more fun, make these a little bit edgier, you know, make these a little bit bolder.’

And so, just giving that input and having it come back and then giving feedback led to getting better results. So the prompt line, you can give it prompts and feedback based on the last work product that it produced for you. And a lot of people don't realize that you don't have to start with a fresh prompt.

You can get more information and more input, and it's pretty impressive because you'll see how more input is required to be successful and to get you there. And also your input needs to be very specific. It's surprising to me how often we use words when we say things, we describe things, but they're just not specific enough for a human to understand, much less for a machine to understand.

I also saw a meme recently that made me laugh. I can't find it. I looked for it yesterday, but it's a good one because there's a fear that these tools are going to put us out of work. And the truth is — and we have a blog post to address this, by the way, that we'll put in the episode notes — you can't take the human out of PR and public relations or depend upon relationships, and machines are just not able to have relationships in the way humans are, and they're not able to have empathy and feelings and conversations and things like humans are. So I don't fear AI at all in our profession. I don't fear AI at all in my company.

If anything, as you can tell, we're trying to embrace it and leverage it and stay ahead of the curve for it so that we can use it to our benefit instead of being afraid of it, or instead of thinking that it's going to replace us. But I'm saying all that also to share that because of that, you know, that it's still not human, right?

We shouldn't fear it. But we should be thinking about the input that we're giving it. So the meme that I saw said, ‘Hey, don't worry. I've been working with clients to create content and creative materials for more than 20 years, and clients never know how to give the input it takes to get to the final result they're looking for.’

And so if they can't give input properly, they're never going to be able to use AI properly. And so, while that was kind of a joke, if you think about it, if your leadership team, your management team, your own internal team aren’t really good at giving input to your in-house or outsourced PR agency, ad agency, design firm or whatever, then they're going to struggle using AI is my point. And so maybe the practice of using AI will help us get our language tighter and more specific when we are giving such input. That's my hope.

So moving on. Let's see. So we've used it to develop product names. We use it to develop themes and meeting names and fun stuff like that, which, by the way, I mean you could go into ChatGPT right now and say, ‘I need 100 names for this product, it does X, Y and Z,’ and within seconds it will give you 100 names. Probably none of them are going to be the one, but you can give it that input to refine, refine, improve. You continue to give it more and more narrow input and expectations and you're going to get something in there that you could at least borrow or inspire something more from.

And that's what I love about it. Again, be careful putting any formal nouns in. Be careful putting anything confidential. Don't violate confidentiality. Don't put something into the system that your client wouldn't want you to. All right. So other examples, we are able to use ChatGPT but I was able to use it to find an article I've been looking for using search engines for about 10 years. Come to find out the article is no longer online, but somehow ChatGPT had archived the article and it was able to summarize the article for me. It wasn't able to reproduce the article in its full because of copyright, it said, which was great, but then I just kept asking it more questions to ultimately get more pulled out from the original article.

That was very helpful to me because like I said, I've been looking for this article for a long time. I think it came out in 2011 and so 12 years later I was able to finally find it after wanting to find it again, and I was able to extract a lot of content out of it that I found very useful.

Speaking of articles, you can take a long article or a report. You can put it into the engine and say, summarize this with a 200-word summary. You can give it the command of too long didn't read, you can give the command of you write this for an audience of a third grader, a fifth grader, a college senior, an entry-level, or a freshman in college kind of thing.

So you can take complicated topics, put them in the system and get some output. Again, output is not going to be perfect, needs to be verified, but it gives you a step in that direction. We also play around with things like telling it, ‘Hey, I need 10 social media posts about this blog post, I need 10 social media posts on this topic,’ and then you input a link to the website or the existing elements.

So for example, we put in a link and said, ‘Here's an Instagram account. Please generate five new Instagram posts for this.’ Or ‘Hey, we need five quick tweets on this particular topic.’ And it did it. We also asked it to say, ‘Hey, what are some suggested SEO keywords for this particular domain or for this product?’ And it came back with terms, most of which we are already doing. But there were one or two that we could take that, tweak this part, and suddenly we might have a new keyword. Let's go do some research on that and see if that would be valuable.

Let's see. Oh, and then finally in closing, the other thing we did is we took a new client that is an executive, a CEO of a company who wants to start speaking to the industry. So we input into the engine. We said, ‘Hey, write a bio about this person.’ That's where the comment I came back earlier saying they kept calling him a senior vice president. He'd never been a senior vice president, he'd been a vice president. And then he got promoted to CEO and we said, ‘Okay, put together using this bio, put together a pitch to have him speak on this topic to this conference with an audience of meeting planners and event organizers.’

So it came back with a nice little pitch and we said, ‘Okay, great. Here's his topic, right? Outline the presentation that he might give at this conference.’ It came back, the outline I thought was fantastic. It was very good. And then we said, ‘Hey, we like this outline. Go ahead and write the presentation for it.’ So it started producing the presentation for it. Ran out of words and then we gave it the command to continue finishing the rest of the presentation. It produced that. It was pretty good, but still very rough.

We would probably make a lot of changes to that, so much so that we didn't even use it. But we will borrow the outline, we will improve upon the outline. We might even input the outline that we revise just to get some ideas for the actual presentation itself. But you still need a professional speechwriter involved. You cannot just depend on this very, very rough robotic computer draft that came forth from it.

From there we decided, ‘Hey, develop a pitch around the presentation both for meeting planners that are going to be selecting and doing a call for presentations.’ It did that, it was pretty good, it was okay. Again, it's going to require a lot of human touch, but it's a head start. And that's the key, is it's a head start.

Then we just started having fun with it. ‘Okay, draft a news release announcing this person speaking at this conference with this topic and presenting on this speech.’ Boom, did it within a matter of seconds, way faster than a human could do it. As good as probably a human could do it, an average human who's not a professional in the industry, but certainly nowhere near as good of quality that our agency would ultimately end up producing this for the client.

Hey, I've been talking a lot here about AI. What other questions do you have about AI? What have been some of your observations? What are your concerns? What are your challenges? How are you using it today? Let's engage in a conversation on social media or otherwise. There'll be a way to engage with me in the episode notes.

But if you like this topic, tell me what other topics you want to hear about. And most importantly, if you enjoyed this episode and you can think of a colleague who would benefit from it as well, please take a minute and share this episode with them. Have some collaboration and communication about it, and let's see how this episode might help you stay on top of PR.